only yielded fruit in four of the five categories. One fruit was then taken from each category
from each orchard for analysis without ripening an 812 fruit sub-sample, and a second for
analysis after ripening 789 fruit; note, some fruit were lost to rots and damage.
On the following day, both sub-samples were measured individually for density by displace-
ment, applying appropriate corrections for ap- paratus volume, fruit and water temperatures,
and the weight of water retained on wet fruit. The non-ripened fruit were also analysed at that
time for DM by removing two 5-mm slices from the equatorial region, weighing them together,
oven-drying at 65°C until constant weight, and expressing the result as a percentage of fresh
weight FW.
Fruit to be ripened were held at 25°C for 2 – 7 days until each reached eating firmness ap-
proximately 8 N, the end point determined by feel and checked using a penetrometer with a
7.9-mm plunger on auxiliary fruit. As each fruit became ripe, it was returned to storage at 0°C,
and when all were ripe, analysis for DM and SSC started within 24 h. Fruit were equilibrated
to room temperature before analysis. Slices 15- mm were removed from each end of the fruit
and juice squeezed from them separately onto a digital refractometer Atago™, Japan calibrated
in °Brix, and the means of the two ends were reported as SSC expressed as °Brix, grams of
sucrose equivalent per 100 g of juice. DM analyses, as described above, were also per-
formed on these fruit.
2
.
2
. Composition trial Kiwifruit were harvested from two Bay of
Plenty orchards in New Zealand during late May 1997, graded and packed as for export and
held in cool-storage 0°C for approximately 2 weeks. One tray of fruit of average size 36
count, 100 g each was randomly selected from each orchard line, and the fruit individu-
ally assessed for density as above. Fruit from each orchard were then ranked by density. Six-
teen pairs of fruit with closely matched densities were selected, four pairs near each of the nomi-
nal densities 1031, 1037, 1044 and 1050 kgm
3
. This provided a sample of matched pairs of
fruit representing the typical range of densities of the New Zealand kiwifruit crop. It is notable
that the eight pairs with the lowest densities came from one orchard and the remainder from
the other. The pairs of fruit were then separated into two sets for composition analysis, one set
immediately, and the other after ripening. This permitted direct comparison of the composition
of pre- and post-ripened fruit with equal densi- ties and therefore similar DMs.
The ‘unripened’ fruit sample was stored at 0°C overnight and analysed the following day
while the ‘ripened’ sample was ripened as above prior to analysis. All fruit were individually
analysed for SSC and DM as described above. SSC readings were completed rapidly to elimi-
nate errors from any starch particles settling on the sensor plate checks on this were made with
repeated readings on duplicate fruit. The re- mainder of each fruit was cut into 2 – 3 mm
slices, weighed, frozen in liquid nitrogen and stored at − 20°C. Later, these samples were
lyophilised, ground and analysed for sugars and starch, expressed as FW Klages et al., 1998.
SSC, as determined by refractometer, is a measure of the soluble material in the juice.
Consequently, it must be adjusted to a whole fruit basis to allow comparison with other fruit
constituent measurements see AOAC method 932.12; Anon, 1995. Using DM and SSC, solu-
ble solids in whole fruit SSF can be calculated as a percentage of FW, using:
SSF FW = SSC100 − DM
100 − SSC 1
3. Results and discussion
3
.
1
. Sur6ey trial The regression of initial density against DM
from the 812 unripened survey fruit Fig. 1 is DM FW = 8.4[ 9 0.13] +
0.210[ 9 0.004] × Density − 1000 2
r
2
= 82.6; S.E.M. DM estimate = 0.68 FW.
Parameter S.E. are shown here and subsequently inside square brackets. The independent variable
‘density − 1000’ is used because water density = 1000 kgm
3
is the notional zero point of the DM scale. It is worth noting that the slope term
0.210 is dependent on the density of the DM of the fruit solids, and the offset 8.4 depends on
both DM density and the amount of gas present Wilson and Lindsay, 1969. If the volume of air
and gases in the fruit were more variable, we would expect the r
2
of the above regression to be smaller.
The regression of initial density against ripe fruit SSC Fig. 2 from 789 ripened survey fruit is
Ripe fruit SSC °Brix = 5.7[ 9 0.11] + 0.189[ 9 0.003] × Density − 1000
3 r
2
= 85.1; S.E.M. SSC estimate = 0.57°Brix.
Because DM was not measured for this sample, it was not possible to calculate the SSF values.
Fig. 2. Ripe fruit SSC vs. initial density and associated regres- sion line of individual kiwifruit sub-sampled from 208 New
Zealand orchards. Error bar indicates one S.E. of the fit of the data to the line.
Fig. 1. DM vs. initial density and associated regression line of individual kiwifruit sub-sampled from 208 New Zealand or-
chards. Error bar indicates one S.E. of the fit of the data to the line.
Considering the extensive nature of the survey, both DM and SSC relationships have high r
2
values, and S.E. reported are likely to be smaller in absolute magnitude than the level that a human
can differentiate.
3
.
2
. Composition trial The means and S.E. of composition measure-
ments of the four fruit in each density category and ripeness group are shown in Table 1.
3
.
2
.
1
. Density and dry matter Density values for the fruit Table 1 covered
the full range of results reported by Asami et al. 1988, and DMs extended from 14 – 19.5 FW
Fig. 3, representing the typical range for ‘Hay- ward’ kiwifruit Beever and Hopkirk, 1990.
Good matches of DM concentrations between the ripenedunripened pairs linked symbols in Fig. 3
were found S.E.M. B 0.25 FW; mean difference
between pairs B 0.1 FW. This pair matching and the DM range produced, justify the use of
density as a selection criterion and created a unique data set with which to explore composi-
tion changes with ripening. Results are clearly grouped within each of the four nominal density
categories Fig. 3, and a trend is evident both between and within the two orchards orchard 1
below and orchard 2 above the 1042 kgm
3
point, respectively. Similar groupings appeared in each
of the other measurements made and showed no suggestion of difficulties caused by taking the fruit
from different orchards. There was no significant difference between the DMs of ripe and unripe
groups.
The regression line for DM against initial den- sity of all fruit shown in Fig. 3 is
DM FW = 8.2[ 9 0.3] + 0.209[ 9 0.008] × Density − 1000
4 r
2
= 96.2; S.E.M. = 0.3 FW. This strong cor-
relation between initial density and DM measure- ments has been reported before Richardson et
al., 1997, however, their coefficients were differ- ent, tending to have lower slopes 0.192, and
higher offsets 10.23 perhaps due to the fact that their measurements were made at normal harvest
time.
Corresponding values for DM obtained by the oven method and for the lyophilised samples
Table 1 matched to 0.6 FW on average, with oven DMs being slightly lower. Only two individ-
ual oven DMs were more than 1.2 FW lower than the lyophilised DMs.
Table 1 Mean and S.E.M. in brackets of measurements made on four fruit in each of four density categories of both unripe and ripe fruit
a
Nominal density group 1044 kgm
3
1037 kgm
3
1031 kgm
3
1050 kgm
3
Unripe fruit 1031.1 0.30
1036.8 0.43 Initial density kgm
3
1044.2 0.42 1049.8 0.68
10.7 0.28 11.2 0.28
11.3 0.21 Initial SSC °Brix
12.3 0.43 11.4 0.43
10.2 0.29 Initial SSF FW
10.6 0.28 10.6 0.24
14.9 0.11 15.8 0.17
17.4 0.18 Initial DM FW
18.6 0.25 Lyophilised DM FW
19.7 0.19 18.0 0.35
16.9 0.22 15.3 0.20
2.6 0.13 2.5 0.13
2.8 0.18 2.3 0.12
Fructose FW 2.1 0.12
2.2 0.11 2.2 0.17
2.4 0.20 Glucose FW
1.4 0.07 1.6 0.07
Sucrose FW 1.3 0.05
1.5 0.08 Minor sugars FW
0.14 B.01 0.14 B.01
0.12B.01 0.13 B.01
2.8 0.52 1.8 0.12
1.2 0.22 3.7 0.54
Starch FW Starch FW
2.0 0.13 1.4 0.25
3.1 0.58 4.1 0.59
Ripened fruit 1031.1 0.32
1036.5 0.42 Initial density kgm
3
1044.1 0.45 1049.9 0.80
1058.3 1.13 1045.9 1.28
Ripe density kgm
3
1052.5 0.99 1040.4 0.51
12.0 0.15 13.3 0.18
Ripe SSC °Brix 14.5 0.21
15.9 0.17 Ripe SSF FW
11.7 0.15 12.9 0.18
14.0 0.23 15.4 0.17
Ripe DM FW 14.5 0.16
16.0 0.19 17.4 0.07
18.7 0.16 15.0 0.10
16.1 0.25 Lyophilised DM FW
17.9 0.06 19.2 0.20
3.7 0.07 3.3 0.06
2.9 0.07 Fructose FW
4.0 0.06 Glucose FW
2.7 0.11 4.0 0.05
3.2 0.10 3.7 0.03
1.3 0.04 1.6 0.06
Sucrose FW 1.6 0.05
1.9 0.07 0.19 0.04
0.24 0.03 Minor sugars FW
0.14 B.01 0.16 B.01
0.02 0.01 0.04 0.01
Starch FW 0.01 B.01
0.02 B.01 0.02 0.01
0.02 B.01 0.01 B.01
Starch FW 0.04 0.01
a
FW, fresh weight; and starch is the starch concentration adjusted by 11 to reflect the amount of sugar it will produce following hydration.
Fig. 3. DM of kiwifruit before and after ripening, plotted against initial density and showing the respective re-
gression lines. Orchard 1 comprises the lower 16 data points. The S.E. of the data about the regression lines is shown as a
bar at the mean density, and paired fruit are shown linked.
Ripe fruit SSF FW = 5.9[ 9 0.47] + 0.188[ 9 0.011] × Density − 1000
5 r
2
= 95.0; S.E.M. = 0.33 FW.
Because ripe and unripe DM levels were matched see Fig. 3, the potential soluble solids
levels of ripened and unripened fruit should not be different. Although the data contributing to
Eq. 5 do not extend across the complete ripening regime from normal harvest to full ripeness, our
own studies Jordan and Seelye, unpublished data and that of Richardson et al. 1997 provide
support for this relationship holding for all ki- wifruit regardless of ripeness status. Initial density
thus provides a means to non-destructively esti- mate the soluble solid levels of ripe kiwifruit. Ripe
fruit SSC can also be estimated from initial den- sity using the following regression
Ripe fruit SSC °Brix = 5.9[ 9 0.43] + 0.198[ 9 0.010] × Density − 1000
6 r
2
= 96.2; S.E.M. = 0.30 FW. This relation-
ship is similar to that reported by Asami et al. 1988 and Richardson et al. 1997 who gave
slopes of 0.171 and 0.165 cf our 0.198 and offsets of 7.89 and 7.47 cf our 5.9, respectively.
It is noteworthy that their relationships pertain to fruit analysed closer to harvest, and, as noted
earlier, show lower slopes and higher offsets.
3
.
2
.
3
. Starch Starch contents of ripened fruit were low in all
cases B 0.05 FW and contributed little to total carbohydrate. Concentrations in unripe fruit
ranged from 1 to 4 FW and were higher in high density fruit. Starch contents were lower than
those typical of fruit at harvest about 7 FW, e.g. Fuke and Matsuoka, 1982; Walton and de
Jong, 1990; Richardson et al., 1997.
By adding the concentrations of starch to solu- ble solids for each unripe fruit, an estimate of the
potential soluble solids levels in ripe fruit can be made. To allow comparison with the sugars, the
starch mass has been scaled up by 11 labelled starch to account for the higher resultant sugar
mass following hydrolysis.
3
.
2
.
2
. Soluble solids The soluble solid concentrations of the ripe
fruit were not dissimilar to the ranges of Beever and Hopkirk 1990, Paterson et al. 1991. SSC
measurements of
unripened fruit
mean, 11.4°Brix were above normal harvest values typ-
ically fruit arrive at the pack house with mean SSCs between 6.5 and 10.5°Brix and comparison
with the fully ripened figures above mean, 13.9°Brix indicated that starch hydrolysis was
already well underway. It was noted that fruit with higher density and DM tended to have
higher SSC values. However, these fruit also had correspondingly more starch showing that starch
conversion was proportionately less advanced.
The regression relationship between initial den- sity and ripe fruit soluble solids expressed as
FW and including S.E. of the estimates is
A comparison
between ripe
and unripe
SSF-plus-starch Figs. 4 and 5 shows regression lines matching to better than 0.4 FW at all
densities, consistent with starch being hydrolysed to sugars on ripening.
3
.
2
.
4
. Sugars Results for ripe fruit sugars Table 1 are typi-
cal of those of Beever and Hopkirk 1990, albeit at the low end of or lower than their ranges, and
fit comfortably inside the ranges of Paterson et al. 1991 except for our highest sucrose value. Wal-
ton and de Jong 1990 report totals of sugars plus starch that are similar, although the relative
fractions of their components differ from ours consistent with the later stage of ripeness of our
fruit.
The major sugars measured had the following concentration ranking, fructose, glucose and su-
crose. Minor sugars myo-inositol, plus a sugar that co-eluted with galactose were also present in
Fig. 5. Ripened closed diamonds and unripened open dia- monds fruit soluble solids plus starch starch scaled by 11,
and unripened fruit soluble solids triangles concentrations of kiwifruit plotted against their initial density with correspond-
ing regression lines. Error bars show one S.E. of the fit to the regression lines and the steepest line is the unripe SSF +
starch.
Fig. 4. DM circles and soluble solids triangles concentra- tions expressed as FW for ripened closed symbols and
unripened open symbols kiwifruit and corresponding regres- sion lines. Error bars show one S.E. of the fit to the regression
lines, and the steeper DM line corresponds to the ripe fruit.
low concentrations 0.17 and B 0.08 FW, re- spectively and were grouped together in this
study. By plotting sugars cumulatively, an illustra- tion of fruit composition and the way it changes
with increased density is provided Fig. 6.
Total carbohydrates in both graphs the top lines in Fig. 6 increased with density, and were
similar in ripe and unripe groups, as planned by matching the original fruit pairs. Sucrose and
minor sugars were largely constant regardless of ripeness-state or DM. While glucose and fructose
in unripened fruit did not increase significantly with density, both showed significant trends with
density in the ripened fruit. Furthermore, the net gains of glucose and fructose on ripening were
nearly
equal, and
together, almost
exactly matched the loss of starch. This follows a pattern
found by Fuke and Matsuoka 1982, Sawanabori and Shimura 1990, and until their fruit became
over-ripe MacRae et al. 1992, and is in agree- ment with glucose and fructose being sourced
from starch.
3
.
2
.
5
. Other soluble components The refractometer-based SSF SSC as FW
measurement, and the chemically determined sug- ars see Fig. 7 show an SSF that is greater than
the sum of sugars by 4 FW, increasing to 5.5 FW in the ripened high density fruit. SSF would
include the three fruit acids citric, quinic and malic which were not measured in these experi-
ments and which are typically 1.0 – 1.6 FW in total Beever and Hopkirk, 1990; Walton and de
Jong, 1990, although as high as 2.9 FW Pater- son et al., 1991. This leaves 2.4 – 3.9 FW cf.
Beever and Hopkirk, 1990, or 1.1 – 2.6 FW cf. Paterson et al., 1991, of the SSF readings uniden-
tified. This discrepancy, which is also apparent in the data of Fuke and Matsuoka 1982, may be
due to the presence of another soluble component that we have not measured e.g. potassium, nitrate
or soluble pectins; see Beever and Hopkirk, 1990. Alternatively, less than the total carbohydrate was
recovered from the fruit samples, resulting in a slight underestimate by our internal standards
adjustments.
3
.
2
.
6
. Other insoluble components For ripened fruit, mean measurements of SSF-
plus-starch over the densityripeness groups were consistently lower than DM by 2.8 – 3.4 FW see
Table 1, while in the unripened fruit, this differ- ence was 3.1 – 3.7 FW. These differences are also
apparent by inspecting Figs. 4 and 5 and repre- sent ‘the insoluble non-starch fraction’ of the
DM.
This insoluble
fraction increases
only slightly with higher DM fruit, and would contain
cell wall, seeds, skin, and minor insoluble compo- nents, and should not be affected by the sugar
and starch levels of the fruit.
3
.
3
. Comparison of sur6ey and composition trial results
The regression lines obtained in the survey and composition trials when compared have very close
parameters for initial density against both DM cf. slope 0.210 vs. 0.209 and offset 8.4 vs. 8.2,
respectively and ripe fruit SSC cf. slope 0.189 vs. 0.198 and offset 5.7 vs. 5.9, respectively. Consid-
Fig. 6. Composition of kiwifruit plotted cumulatively in terms of sugars and starch starch scaled by 11 in unripened and ripened fruit showing regression lines. Error bars show one S.E. of the regression fit to cumulative carbohydrate measurements. Note: upper
symbols in ripened fruit plot almost coincide due to the low starch concentrations.
Fig. 7. Soluble solids in whole kiwifruit SSF, and total sugar estimates for unripened and ripened data groups plotted against initial fruit density and showing regression lines. Error bars show one S.E. of the regression fit to the cumulative carbohydrate
measurements.
ering the extensive nature of the sample taken in this survey, this confirms the robustness of the
relationship in kiwifruit between initial density and the fruit DM and ripe fruit SSC.
It is worth noting that the sets of fruit in the survey and composition trials were stored for
similar lengths of time prior to analysis and would have been at similar stages of ripeness.
Had these trials been undertaken immediately after harvest, the regression parameters would
have been different from those reported above, but would still have matched each other closely.
This predicted change in regression parameters as fruit ripened might be due to a reduction of
internal air spaces during early storage, but war- rants further investigation.
4. Conclusions
